Calrec 1061 Filter Simulation - help needed.

[TomWaterman]

Hi folks, trying to get my head around this Calrec 1061 project. Currently looking at the Mid EQ and the associated frequency curves for the original inductor/cap values plus new ones I have calculated (for available Wima's (MKS2 / FKP2) and a Sowter 9312 ('Neve' T1530) inductor).

Here is the original schematic.

I think it would be cool to use an available off the shelf inductor like the Sowter so that its always available for DIY - no hard to get custom ordered parts. Custom Sowter inductors would cost more than the 9312 so I'd like to work with this.

I spent some time calculating all of the freq values for the Wimas and each tap on the Sowter (2, 1.1 and 0.45H). I'm also reducing the number of available freqs to 5 plus off (instead of 10) as this means Lorlins can be used, keeping the cost down.

I have picked 5 freqs from my chart which cover most areas and are spaced roughly in line with a Neve 1073 and Wunder PEQ1:

360Hz, 760Hz, 1k5, 3k2 (or 3k4) and 6k1.

These freqs also remove multiple series caps (found in the original), reducing parts count. Anyway I wanted to see how the new values would alter the bandwidth of the filters compared to the original so I've tried simulating it.

Never done this before so I was wondering if anyone can comment on my results??

Here are the original curves from the data sheet.

Here is my simulated circuit for the MF in Lo-Q.

Here is my simulated circuit for the MF in Hi-Q.

How do these look to you guys? I'm not certain about the loading of the filter section and how that should relate to the next amplifier stage, however my results look as I'd expect.

Here are my simulated curves for the original values @ Lo-Q (the original frequencies are 270, 390, 560, 820, 1k2, 1k8, 2k7, 3k9, 5k6 and 8k2):

Here are my simulated curves for the original values @ Hi-Q:


They look pretty good to me. Any thoughts, I'm a bit new to this simulator thing?

So here are the new freqs that I simulated. Graphs show that the bandwidth has narrowed compared to the original but its the same order of magnitude so to speak. The 360Hz (lowest freq) is maybe a bit narrow and the 3k2 is maybe a bit too wide compared to the original but I like that, nice smooth wide curve for vox lift, snare etc. In Hi-Q mode the BW is pretty close to the original, just a bit tighter - any thoughts? I'd like to hear from Cannikin or Buz about the sound of the original in Hi-Q mode???? how is it guys?

New values @ Lo-Q (just noticed the AC sweep didnt go high enuff on these, deleted the good one..bummer.):

New values @ Hi-Q:

I tried altering the Q resistor to see if I could broaden the BW but its only very slight before it reduces the available range of boost.

The ranges shown in the graphs above are pretty much spot on compared to the module specs, which is +/-16dB in Hi-Q and +/-12dB in Lo-Q.

Here is a phase and freq plot of the 360Hz setting with the original 18k BW resistor @ both Q settings.

And here is a phase and freq plot with an adjusted BW resistor (10k) @ both Q settings.

Right my question is, how do I set up the simulator to generate boost graphs? There appears to be something funny going on with the gain pot (25k lin). The following graph shows the pot in 10 dif. positions from 0% to 100% in 10% steps but its don't look so right to me! (Reference points on the Y-axis that is)....

From the schematic 50% should be no cut or boost, 0% should be max boost and 100% should be max cut. Its obvious to me its something to do with the way the output of the amplifier stage feeds back into the network via a 3k5 resistor to the other side of the gain pot.

[EDIT] Just realised I didn't mention that 0% is the flat line and 100% is max cut here.[/EDIT]

Can anyone explain to a dummy like me whats actually happening here, I'd like to know how to simulate it?

How does the stage generate cut and boost?

Any help would be mucho appreciated.

Cheers Tom

 

[TomWaterman]

Bump before it falls over the page.

Anyone got anything to add, it would be greatly appreciated?

Thanks a lot
Tom

 

[davebungo]

I think you really need to include the Op-amp (I assume that's what it is) in the analysis.

 

[TomWaterman]

Thanks Dave, thats what I've been thinking.

Its a discrete amp and I suppose it is integral to the filter and its operation. I only have limited transistor models in SImetrix and I'm also very new to the simulation thing.

I'll see if I can get away with basic transistor models this week.

Thanks for the thoughts. No one got any other thoughts on the curves as they are now, especially with regards to the new choices and the changing bandwidth??

Tom

 

[soundguy]

hi tom-

crazy work. My first thought halfway through reading your post was that it would be cool to make the curve wider, not narrower and then I finished reading the post and got my answer to that response.

Its really cool to see this one progressing forward.

dave

 

[PRR]

> how do I set up the simulator to generate boost graphs?

You need the amplifier to make this puppy boost.

It does not have to be THAT amplifier; if I read it right, you could use a generic GAIN block with gain set to about -1,000. The advantage of GAIN over OPAMP is that (at least in my sim) you don't need power supply and it never clips.

You can leave out the 10Meg resistors: they are just there to bleed the clicks out of the switched caps, which is irrelevant in a sim.

The 100UFd cap is part of the DC biasing for what seems to be a single-supply power situation, with an ideal gain block you can replace that cap with a short.

 

[PRR]

> I tried altering the Q resistor to see if I could broaden the BW

The Q, input/feedback, and pot resistances set the boost/cut.

To change the Q with the same resistors (same boost/cut), change the L and C in opposite directions.

 

[TomWaterman]

Hi again,

Many thanks PRR for the simple clarification. I just used a TLO72 model and got it running in simulation nicely indeed. I did use appropriate load resistors etc to ensure I got close to the original. Its my understanding now that the gain pot controls the feedback around the amplifier along with the filter network. I've played quite a bit in the simulator to see how everything interacts.

Dave are you saying that the originals would sound subjectively 'better' (or be more useful) with a wider BW?

After simulation, apart from the lowest of my proposed new frequencies, they are pretty wide IMO.

Here is my HTML version of the spreadsheet for all MF freq values with each Wima cap. So if there are any freqs you would rather have, I can simulate them for you. Like I said earlier I tried to pick an even spacing much like a 1073. As a general rule, once you get to about 1kHz, go for the 1.1 or 1.2H values and at about >4kHz go for the 0.45 or 0.7H values. That will get you close to the original curves (0.7H will be wider).

My first pick was (in Hz) 360, 760, 1k5, 3k4 and 6k1, which looks like this:

I then did some for the other Sowter inductor (9805) which is 2, 1.2 and 0.7H - and should lead to wider curves (at the 1.2 and 0.7H taps).

This one was (in Hz) 360, 760, 1k4, 3k3 and 6k, which looks like this:

The 3k3 is a fair bit wider than the 3k4 in the first and the 6k is fookin' huge - too much I think. I think I prefer the first ones.

Here is another one to see a few different freqs. If you guys are worried that the 360Hz is too narrow then you could start at around 520Hz (like the Avedis E15) and get smoother transitions.

I think both inductors have the same pin-out, so hopefully you can have somewhat 'custom' curves by picking your own values.

I played with the 'Q' resistor (which is normally 18k in Lo-Q mode on the original) and have settled upon two resistor values that can be switched between instead of switching the resistor out to achieve Hi-Q as in the original.

By using 47k as the 'Q' resistor, the curves are very similar to Hi-Q in the original and it also limits the range to +/-15dB. This EQ is like some Neve EQs I think. Hi-Q mode does nothing but add more gain range (from +/-12db to +/-16dB in the original). If you back off the gain in Hi-Q mode you pretty much land on the same curve as Lo-Q.

I found that another 'Q' resistor value of 8k6 was pretty cool. It limits the range to +/-9dB but actually does make the curve broader and less 'peaky'. My suggestion is to toggle between 8k6 and 47k for either +/-9dB or +/-15dB of MF EQ. the 8k6 mode should be really nice for smooth vocal lifts etc. The following graph shows that the 8k6 shape cannot be matched by backing off the 47k curve. Its subtle but more of a substantial difference than the original.

The following two plots show the 'Q' resistors effect @ 360Hz and 3k2:

Next two show their effects on max cut & boost for the five mid positions:

Narrow BW

Wide BW

What do ya reckon?

Here are plots for each individual mid position - I think this kind of filter is called "Proportional Q"??

360Hz
760Hz
1k5
3k4
6k1

Any thoughts?

I also plotted the LF shelf curves:

Which seem fine. Here is the cut & boost range at 100Hz - looks like the filter adds a small resonant peak near the pass band.

I did the HF shelf too, but it seems a bit weird, there is a broadband loss or gain of about 0.5-1dB depending upon position - is this normal? The full original schematic can be found here. I also noticed what I think is a schematic error too, C9 and C14 in the HF should be the same value, not 12n3 and 3n3. I set them both to 3n3 for the sim.

And here for the cut & boost range @ 6k8 shelf...

OK so I guess I'll redraw the EQ schematic next for a 6 pos switch and remove the HPF and LPF filters. I'd like to talk a bit about caps with you guys. It looks like the original uses Polystyrene and Polyester, how is it with you if I lay out boards for Wima Polyesters and Props?

I think they will be available for a long time to come and they sound good apparently!

Let me know your thoughts on practical mid-frequencies. We all have different uses.

Cheers Tom